WPS5642 Policy Research Working Paper 5642 Stimulating Managerial Capital in Emerging Markets The Impact of Business and Financial Literacy for Young Entrepreneurs Miriam Bruhn Bilal Zia The World Bank Development Research Group Finance and Private Sector Development Team April 2011 Policy Research Working Paper 5642 Abstract Identifying the determinants of entrepreneurship is Herzegovina. The authors conduct a randomized control an important research and policy goal, especially in trial and find that while the training program did not emerging market economies where lack of capital influence business survival, it significantly improved and supporting infrastructure often imposes stringent business practices, investments, and loan terms for constraints on business growth. This paper studies the surviving businesses. Entrepreneurs with higher ex‐ante impact of a comprehensive business and financial literacy financial literacy further exhibited some improvements in program on firm outcomes of young entrepreneurs business performance and sales. in an emerging post‐conflict economy, Bosnia and This paper is a product of the Finance and Private Sector Development Team, Development Research Group. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The authors may be contacted at mbruhn@worldbank.org and bzia@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Stimulating Managerial Capital in Emerging Markets:    The Impact of Business and Financial Literacy for Young Entrepreneurs    1 Miriam Bruhn and Bilal Zia                                                                            1 Both authors are from the Finance and Private Sector Development Team of the Development Research Group at  the World Bank. We would like to thank the World Bank Group and the Ewing Marion Kauffman Foundation for  financial support, and Partner Microcredit Foundation, in particular Selma Cilimkovic and Selma Jahic, for sharing  their data with us and for their outstanding collaboration and support throughout the project. We are also grateful  to  Fenella  Carpena,  Sabina  Djonlagic,  and  Adnan  Mesic  for  providing  excellent  research  assistance  and  to  David  McKenzie and conference participants at the World Bank for helpful comments.  I. Introduction  Much of the literature on the determinants of entrepreneurship and firm growth has focused  on access to physical capital and external finance (e.g. Banerjee et al., 2010; Bruhn and Love,  2009; and De Mel, McKenzie and Woodruff, 2008). However, a number of recent papers argue  that “managerial capital” or business skills are another important driver of firm growth and a  key determinant of productivity (e.g. Bloom et al, 2010; Bruhn, Karlan, and Schoar, 2010).     This  emerging  academic  interest  in  identifying  alternate  channels  of  firm  growth  has  been  accompanied  by  an  equally  strong  policy  interest  in  education  programs  geared  towards  enhancing  financial  and  business  skills.  Governments  and  private  organizations  alike  are  investing  heavily  in  financial  literacy  programs  throughout  the  world. 2   Despite  this  attention,  we know very little about what kinds of education programs are effective and for whom.         For example, the only completed randomized evaluation of a financial literacy training program  designed  to  promote  savings  behavior  (Cole,  Sampson,  and  Zia,  2010)  finds  no  effect  of  the  training  on  the  overall  population  in  Indonesia,  though  it  does  find  a  small  increase  in  the  probability  that  individuals  with  low  initial  levels  of  financial  literacy  open  bank  accounts  following the training.                                                                2   See,  for  example,  Cole  and  Fernando  (2008),  or  http://corporate.visa.com/viewpoints/responsible‐ spending/financial‐literacy.shtml  2      The  evidence  on  the  effects  of  business  training  on  entrepreneurial  outcomes  is  also  scarce.  Karlan  and  Valdivia  (2010)  find  that  a  business  education  program  for  female  micro‐ entrepreneurs  in  Peru  improves  record‐keeping,  though  not  profits;  and  Drexler,  Fischer  and  Schoar (2010) show that a basic rules‐of‐thumb based training, but not formal business training,  leads  to  improvements  in  business  outcomes  for  micro‐entrepreneurs  in  the  Dominican  Republic.     Our  paper  adds  to  the  sparse  knowledge  base  on  the  effects  of  business  and  financial  education.  We  focus  on  young  borrowers  in  Bosina  and  Herzegovina  who  are  business  loan  clients of our partner financial institution, Partner Microcredit Foundation (henceforth Partner),  operating  within  and  near  the  metropolitan  city  of  Tuzla.  Bosina  and  Herzegovina  is  an  important  location  choice  since  it  represents  an  emerging  post‐conflict  economy,  struggling  with the burden of high youth unemployment and low business survival. In such a setting, the  marginal value of a business and financial education program is likely very high.      At  the  time  of  the  baseline  survey,  approximately  one‐third  of  our  sample  did  not  own  a  business but had a business exploration loan with Partner. These sample features enable us to  makes three important contributions to the literature: (i) we study the effects of financial and  business training not only on existing business owners, but also on potential entrepreneurs to  3    identify  impacts  on  business  startup;  (ii)  we  focus  on  slightly  larger  businesses  than  micro‐ enterprises,  some  of  which  have  employees,  own  operational  assets,  make  business  investments,  and  are  formally  registered;  and  (iii)  we  utilize  very  detailed  and  high  quality  administrative  loan  data  to  study  impacts  on  default  rates  and  loan  terms,  in  addition  to  analyzing survey measures on business outcomes.        Our  research  design  is  a  randomized  control  trial  with  445  Partner  loan  clients,  two‐thirds  of  whom  received  an  invitation  to  attend  a  comprehensive  business  and  financial  education  program run by a highly experienced and reputable training institute in the city. The remaining  one‐third  of  the  sample  is  our  control  group.  The  randomization  was  stratified  by  baseline  financial literacy level, gender, industry, and baseline profits.     We find that financial literacy is a strong predictor of baseline financial and business outcomes,  consistent with the existing literature. Further, our experimental results show that the training  program led to significant improvements in basic financial knowledge for those who start out  with low levels of financial literacy at baseline.     Our  results  on  business  outcomes,  on  the  other  hand,  are  quite  stark.  We  do  not  find  any  significant treatment effects on the extensive margin. Specifically, treatment businesses were  no  more  likely  to  survive  than  control  businesses  in  a  period  where  36  percent  of  businesses  4    shut down by the time of the follow‐up. 3  In addition, we find no significant treatment effect on  business start‐up, with only one new business starting up during our study period. These results  clearly  suggest  that  lack  of  business  acumen  is  not  the  primary  driver  of  business  entry  and  survival.        While the extensive margin results are not significant, we identify positive treatment effects for  businesses  that  remain  operational  during  the  study  period.  The  strongest  effects  are  on  improvements in business practices and investments. We find that our treatment group was 17  percent  more  likely  to  implement  new  production  processes  than  the  control  group,  and  11  percent  more  likely  to  inject  new  investment  into  the  business.  These  results  are  consistent  with  the  central  theme  of  our  business  training  program,  which  was  to  encourage  capital  investment  among  young  businesses.  Further,  those  invited  to  the  business  and  financial  training were substantially more likely to separate their business and personal accounts, which  was also emphasized in the business training course.     In terms of business performance, while we do not find significant average treatment effects of  our training program, we identify significant heterogeneous effects. Specifically, entrepreneurs  with  high  ex‐ante  financial  literacy  exhibit  significantly  greater  improvements  in  sales  due  to                                                               3  The large proportion of firms shutting down is consistent with Demirguc‐Kunt, et. al. (2007), who find that nearly  50 percent of new businesses in Bosina and Herzegovina do not survive beyond their first year.   5    the  training  program  than  entrepreneurs  with  low  ex‐ante  financial  literacy.  The  effects  on  profits are also positive for this sub‐group, showing an increase in profits due to the training of  54 percent, though only statistically significant at the 15 percent level.        Next, we study treatment effects on external finance. We use detailed administrative data from  Partner to study the effect of the training on default rates, propensity to refinance, and terms  for new loans. Although we do not identify average treatment effects on default rates, we find  that  our  treatment  group  was  significantly  more  likely  to  refinance  its  existing  loans  with  Partner.  This  restructuring  can  take  the  form  of  a  lower  interest  rate  or  longer  loan  term.  Further,  we  find  treatment  firms  that  take  on  new  loans  were  significantly  more  likely  to  negotiate  larger  number  of  installments,  controlling  for  loan  amount.  These  results  are  again  consistent with the notion of larger up‐front capital investment, a concept that was central to  our business training program.     Overall,  our  results  have  important  policy  implications  for  business  promotion  and  growth.  First, our results clearly indicate that lack of business acumen is not the primary constraint to  business  survival.  Hence,  business  training  programs  alone  are  likely  not  the  panacea  for  promoting  new  business  growth  in  emerging  markets.  Our  second  set  of  results,  however,  shows  that  business  training  indeed  is  a  strong  complement  to  achieving  such  growth.  In  particular, we find business training can provide the necessary motivation and entrepreneurial  6    impetus  for  existing  businesses  to  grow.  Further,  our  analysis  identifies  specific  business  decisions  for  which  financial  education  for  entrepreneurs  can  be  particularly  effective.  These  insights  are  very  helpful  for  formulating  and  adjusting  policy  advice  so  limited  development  resources can be effectively targeted.      This  paper  proceeds  as  following.  Section  II  describes  the  setting  and  sample  selection,  and  Section  III  outlines  the  research  design  and  summarizes  the  business  and  financial  literacy  program.  More  details  on  the  program  are  provided  in  Appendix  1.  Section  IV  describes  the  implementation  challenges  we  faced  and  provides  summary  statistics.  Section  V  presents  the  baseline analysis, as well as the evaluation results. Section VI concludes.      II. Setting and Sample Selection  For  the  implementation  of  this  study,  we  partnered  with  one  of  the  largest  microcredit  institutions in Bosnia, Partner Microcredit Foundation. 4  Unlike typical microfinance institutions  that cater to the poorest segments of the population, Partner regularly makes large loans, all on  an individual basis and with full credit checks.    All  participants  in  our  study  are  Partner’s  loan  clients.  In  order  to  select  our  study  sample,  Partner  provided  us  with  a  list  of  their  active  borrowers  between  the  ages  of  18  and  35.  We                                                               4 Partner had close to 55,000 active borrowers in 2009.  7    chose loan clients in this age bracket because Partner felt that business and financial education  could  have  a  particularly  large  impact  on  this  group.  Youth  unemployment  is  high  in  Bosnia,  about  58  percent according  to  the  2007  Labor  Force  Survey,  and  self‐employment provides  a  viable  solution  to  this  problem.  In  this  type  of  environment,  it  is  particularly  important  to  explore strategies to promote the entry, survival, and growth of youth‐led businesses.    We  limited  our  study  sample  geographically  to  areas  around  Tuzla, 5   where  Partner  is  headquartered, to facilitate the logistics of the business training. Moreover, we dropped clients  who had not taken out a loan for business purposes from our sample in order to target clients  who  were  either  running  a  business  or  planning  to  start  a  business.  We  also  did  not  include  clients who were delinquent on their loan payments according to Partner’s definition. 6    All 2,274 Partner clients meeting these criteria received an initial screening phone call, asking  them whether they would be interested in participating in a business and financial education  training  course.  About  500  clients  could  not  be  reached  over  the  phone.  Among  the  1,783  clients who were reached, half reported being interested in participating in the course. Table 1                                                               5 We  limited  the  sample  to  clients  living  in  the  municipalities  of  Banovici,  Gracanica,  Gradacac,  Kalesija,  Lukavac,  Sebrenik, Tuzla, and Živinice.  6  Partner’s definition of delinquent loans is either being more than 15 days late on the current payment or having a  cumulative number of late payment days over 15. The reason for not including these clients in the sample is that it  is Partner’s policy not to offer any programs or new loans to delinquent clients. 8    examines which borrower and loan characteristics predict whether the client was interested in  the course. These characteristics all come from Partner’s client database.    Table  1  shows  three  specifications,  one  with  demographic  characteristics  alone,  the  next  one  adding loan characteristics, and the final one with Partner branch fixed effects. All specifications  show  that  women  were  about  13  percent  less  likely  to  be  interested  in  participating  in  the  training. 7   In addition, clients who had been late on at least one of their loan payments (during  the  course  of  the  loan)  were  5  percent  more  likely  to  be  interested  in  training.  On  average,  almost  60  percent  of  clients  had  made  a loan payment  at least  one  day  late,  but the  median  number  of  days  late  was  relatively  small  (i.e.  two  days).  This  last  result  is  promising  in  that  people who were late on payments perceive business and financial education as being valuable.     None of the other variables show a statistically significant correlation with being interested in  training.  Most  notably,  neither  the  client’s  age,  nor  the  loan  amount,  predicts  whether  the  client is interested in participating in the business and financial education course.    In our study, we only include clients who were interested in the training. This implies that we  measure  the  impact  of  training  only  on  the  population  of  interested  clients.  For  policy                                                               7  About 35 percent of the clients who met all selection criteria were women.   9    purposes, this is probably the most relevant sample since only clients who are interested in the  training will take it up if offered the training.    III. Curriculum Details and Research Design  III.I. Curriculum Details  The  business  training  was  provided  through  a  local  NGO,  the  Entrepreneurship  Development  Center (EDC). EDC is located on the premises of the Chamber of Commerce of Tuzla Canton and  has extensive experience with providing entrepreneurship training to university students. Most  of EDC’s instructors are faculty members at the University of Tuzla.     For the purposes of our study, EDC adapted its regular business training course curriculum to  meet  the  needs  of  our  target  audience.  In  order  to  do  this,  they  conducted  face‐to‐face  interviews  with  existing  Partner  loan  clients  and  consulted  with  Partner’s  credit  officers  in  various  field  offices  in  the  Tuzla  region.  Moreover,  EDC  pilot  tested  the  new  curriculum  with  first  year  university  students  who  resembled  our  target  group  in  terms  of  age,  previous  education, and income.    The business training offered through our study consists of six comprehensive modules. These  modules introduce basic business concepts and accounting skills, such as separation of business  and  personal  household  accounts,  and  they  also  explore  deeper  concepts  such  as  business  10    investment  and  growth  strategies.  The  advantages  of  up‐front  capital  investment  are  particularly  highlighted  throughout  the  course.  Appendix  1  includes  a  detailed  description  of  the topics covered in each module.      As part of our implementation strategy, we hired two local consultants to handle the logistics of  the  business  training,  including  calling  Partner’s  clients  and  scheduling  them  for  make‐up  sessions  in  case  a  session  was  missed.  The  training  was  typically  held  in  groups  of  six  to  ten  clients.  The  consultants  also  kept  track  of  attendance,  administered  a  short  follow‐up  test  at  completion  of  the  course,  collected  course  evaluation  forms,  and  distributed  certificates  for  completing the course. Clients were paid 50 KM (approximately US$35) for participating in the  course in order to compensate them for the opportunity cost of their time 8 .    III.II. Research Design  Our research design is a randomized control trial with a sample size of 445 active business loan  clients 9 .  We  originally  envisioned  two  distinct  treatment  groups,  one  receiving  the  first  five  modules  of  the  business  training  course,  and  the  other  an  additional  module  on  issues  pertaining  to  the  financial  crisis.  149  clients  were  randomly  allocated  into  treatment  group  1                                                               8  We also offered clients free of charge transportation to the training location.  9  These 445 are a subgroup of the clients who said that they were interested in the training in our screening phone  calls. We provided the list of interested clients to the survey firm for the baseline survey and asked them to stop  surveying after they had completed 450 interviews. For various reasons, we only ended up with 445 valid baseline  interviews, which form the sample for our experiment.  11    and 148 clients were randomly allocated into treatment group 2, while the remaining 148 acted  as the control group.      We performed a stratified randomization, using information from Partner’s database and from  a baseline survey conducted in April and May 2009. In the baseline, we collected information  on  measures  of  financial  and  business  knowledge,  education,  and  risk  aversion,  as  well  as  business employment, assets, expenditures, sales, profits, and use of external finance.     The randomization was stratified by gender, sector (Farming & Livestock, Services, and other),  above and below the median of the business knowledge/financial literacy score in the baseline  questions, and a dummy for whether profits were missing in the data. Within strata, we sorted  by baseline profits and randomly allocated clients to our three experimental groups within each  sequence of three observations.    The implementation of the business training was carried out soon after the baseline, between  June  and  December  2009.  An  exit  test  to  measure  business  and  financial  knowledge  was  administered at the end of the training to all participants. Finally, a telephone‐based follow‐up  survey was conducted in May and June 2010, one year after the baseline survey. For the follow‐ up, we were able to track down and interview 396 out of the 445 individuals in our study. The  attrition rate was relatively low, and uncorrelated with our treatment.    12      IV. Implementation Challenges and Summary Statistics  The  implementation  of  the  business  training  program  was  quite  challenging.  We  faced  considerable  reluctance  from  our  treatment  group  for  attending  the  course,  despite  the  fact  that  our  entire  sample  consisted  of  individuals  who  had  initially  expressed  interest  in  such  a  course. Out of 297 individuals in the treatment group, only 117 (39 percent) actually attended  the course.     In  the  follow‐up  survey,  we  asked  for  the  main  reason  why  treatment  individuals  did  not  participate in the training program, and the overwhelming reason was lack of time. However,  among  the  people  who  did  attend,  the  satisfaction  rate  was  quite  high,  with  more  than  96  percent of people agreeing that they would recommend this course to a friend.     Given our low attendance figures, and the fact that only a handful of individuals in the second  treatment  actually  attended  the  sixth  module,  we  decided  to  forego  our  original  experiment  design of two separate treatment groups, and merged both treatment groups into one.         Yet another complication we faced was that not all of the 445 clients in our sample actually had  a business at baseline, even though they had a business loan at that time. We were not aware  of this at the time we were designing the experiment protocols, and only later did we identify  13    that  about  one‐third  of  our  baseline  clients  did  not  have  an  operating  business  at  baseline.  Partner  later  explained  to  us  that  these  clients  most  likely  received  the  business  loans  for  a  planned or potential business venture. While on the one hand, we were unable to stratify on  this variable, on the other hand this variation in the sample offers us the opportunity to study  the impact of business training on new business start‐up. Indeed, potential entrepreneurs are  likely prime candidates for whom business training would be beneficial.     From a sample composition point of view, our treatment group is not unbalanced in terms of  individuals  who  did  or  did  not  have  a  business  at  baseline.  In  fact,  there  are  no  statistical  differences between the ratio of treatment and control samples for these two groups. Further,  the  business  training  attendance  data  shows  that  individuals  with  and  without  businesses  at  baseline were equally likely to attend and complete the course, with a mean attendance rate of  39.4 percent and 39.3 percent, respectively.         Table  2  provides  summary  statistics  for  the  baseline  survey,  broken  down  by  treatment  and  control groups. The last column provides p‐values for a difference‐in‐means test between the  two groups. Panel A presents a summary of demographic and stratification variables, and Panel  B focuses on business characteristics for those individuals who had a business at baseline. The  businesses  in  our  sample  have  about  two  employees  on  average  (including  the  owner)  and  14    monthly profits of around KM 1,000 (US$700). They are about 5 years old and 20‐30 percent of  them are registered with the authorities.     Overall, the means of the baseline variables are very similar across the treatment and control  groups.  In  particular,  none  of  the  stratification  variables  are  significantly  different  in  the  sub‐ sample of business owners. There are only a few exceptions, most notably among the business  variables. However, these differences are entirely due to chance. We can be sure of this since  we performed the randomization ourselves. Following the suggestions in Bruhn and McKenzie  (2009),  we  control  for  strata  dummies  and  also  for  baseline  outcome  levels  in  our  regression  analysis.    V. Analysis  V.I. Baseline Analysis of Financial and Business Knowledge  Our  baseline  survey  measures  business  and  financial  knowledge  through  eight  questions  that  are listed in Appendix 2. We construct an overall business and financial literacy score by tallying  the  correct  answers  to  these  eight  questions.  The  score  thus  runs  from  zero  to  eight.  The  average of this score is about 2.7, meaning that, on average, clients gave the correct answer to  2.7 out of 8 questions.    15    Table 3 studies the determinants of formal financial services usage and business practices. The  first  column  for  each  dependent  variable  includes  the  full  sample  and  the  second  column  focuses exclusively on individuals who had a business at baseline. Consistent with the existing  literature from developed and developing countries (e.g. Lusardi and Tufano (2008) in the US;  Cole, Sampson, and Zia (2010) in Indonesia and India; and Klapper and Lusardi (2010) in Russia),  we  find  that  business  and  financial  knowledge  is  a  strong  predictor  of  usage  of  financial  services, including having a bank account and a credit line. Further, entrepreneurs with higher  business and financial literacy are more likely to use trade credit and to keep business accounts.     Apart  from  business  and  financial  literacy,  we  find  that  being  formally  registered,  having  participated  in  a  business  training  program  in  the  past,  and  having  business  assets  are  significant  predictors  of  financial  services  usage.  These  results  are  consistent  with  standard  models of firm behavior as entrepreneurs with more experience and who operate larger firms  are likely to interact more with the formal financial system.      V.II. Predictors of Take Up  As mentioned above, 39% of the individuals who were invited to the business training program  actually  attended.  Table  4  presents  results  of  regressing  attendance  on  various  baseline  characteristics  of  those  invited.  We  find  that  individuals  in  rural  areas  were  significantly  less  likely  to  attend  training,  even  though  all  participants  were  compensated  for  their  travel.  16    Perhaps the greater distance and time of travel imposed restrictions on their attendance. There  are  some  significant  differences  by  ethnicity,  but  more  than  95%  of  the  sample  was  Bosniak,  and hence this represents only a small difference in real terms. Importantly, we do not find any  significant differences in attendance rates by baseline levels of financial literacy, schooling and  age. These results are similar within the sample of individuals who had a business at baseline  (Column 2).       V.III. Evaluation Specification  Since  treatment  was  randomly  assigned,  we  estimate  causal  impacts  with  the  following  equation:  yi = α + β * TrainingIn vitei + ε i                 (1)     where the dependent variable is the knowledge, business performance, or loan behavior metric  used in the regressions. The main coefficient of interest is  β ,  which represents the treatment  effect  of  being  invited  to  our  business  and  financial  education  program.  We  focus  on  the  reduced‐form relationship because it is difficult to compel people to attend a training session;  thus, the intention‐to‐treat estimate may be of greatest interest.     17    Whenever  available,  we  follow  the  recommendation  in  McKenzie  (2011)  and  control  for  the  baseline  value  of  our  dependent  variable  and  run  an  Analysis  of  Covariance  (ANCOVA)  specification. In addition, all specifications include strata dummies and a survey wave dummy  since our follow‐up survey was conducted over two waves. 10        V.IV. Evaluation Results – Effects on Business and Financial Knowledge and Perceptions  In  order  to  assess  the  effect  of  the  training  on  business  and  financial  knowledge,  we  first  examine the results from the exit test that participants filled out at the end of the training. This  test includes the same eight business and financial knowledge questions as the baseline survey.  Results from this exit test are only available for entrepreneurs who attended the training and  thus  cannot  be  compared  to  a  randomly  chosen  control  group.  However,  comparing  exit  test  results  to  baseline  answers  provides  a  first  indication  of  whether  participants  improved  their  business and financial knowledge after the training.     Panel A in Table 5 shows the fraction of respondents who answered each question correctly, at  baseline  and  during  the  exit  test.    The  fraction  of  correct  answers  during  the  exit  test  is  significantly higher than at baseline for three out of the eight questions. Somewhat surprisingly,  respondents also did significantly worse during the exit test in answering two out of the eight                                                               10   The  second  wave  was  necessary  as  the  response  rate  was  initially  very  low.  This  initial  non‐response  is  not  correlated with treatment.   18    questions, compared to the baseline. However, the total score (i.e. the sum of correct answers  on the eight questions) increased significantly from 2.6 to 2.9 after the training, suggesting that  the training improved business and financial knowledge on average.    The baseline and exit test also included a number of questions to measure financial perception  and  attitudes,  such  as  risk  aversion  and  preference  for  using  credit  vs.  own  funds  to  finance  purchases.  Panel  A  in  Table  6  illustrates  that  financial  perceptions  changed  significantly  from  baseline to the exit test. Specifically, respondents were more risk averse after the training and  less  likely  to  prefer  using  credit  instead  of  own  funds.  Moreover,  respondents  had  a  better  understand of the importance of having a good credit history. In fact, before the training only  22 percent of entrepreneurs thought that a good credit history could help them obtain larger or  better loans, while 75 percent of entrepreneurs thought so after the training.     Our follow‐up survey was conducted over the phone, and therefore did not allow us to ask all of  the business and financial knowledge and financial perception questions. Instead, we chose to  include  only  the  three  shortest  and  easiest  to  administer  business  and  financial  knowledge  questions  in  the  follow‐up  survey.  As  shown  in  Table  5,  these  questions  test  whether  the  respondents know VAT law, whether they know what the credit registry is, and whether they  understand  diversification.  The  results  in  the  last  column  of  Table  4  indicate  that  all  training  participants were significantly more likely to answer these questions correctly at follow‐up than  19    at  baseline.  However,  entrepreneurs  in  the  treatment  group  who  did  not  participate  in  the  training, as well as entrepreneurs in the control group also did better at answering two of these  questions at follow‐up than at baseline.     In Table 7, we thus turn to estimating the causal impact of the training on business and financial  knowledge, using the specification described in Section V.III. Here, our measure of business and  financial knowledge is the sum of correct answers to the three questions that were included in  the follow‐up survey, as explained in the previous paragraph. The result in Column 1 indicates  that  the  average  treatment  effect  of  the  training  on  business  and  financial  knowledge  is  positive,  but  not  statistically  significant.  We  then  examine  whether  this  treatment  effect  differed by whether the entrepreneur had a baseline financial literacy level above or below the  median. Column 2 shows that the effect of the training on business and financial knowledge is  positive  and  statistically  significant  for  individuals  with  below  median  financial  literacy  at  baseline.  For  these  individuals,  the  training  increased  the  business  and  financial  knowledge  score by 0.239 compared to the control group mean of 0.897. On the other hand, for individuals  with above the median financial literacy at baseline, the training appears to have had no effect  on  our  measure  of  business  and  financial  knowledge.  This  does  not  necessarily  imply  that  individuals  with  above  median  financial  literacy  at  baseline  did  not  learn  anything  in  the  training since the course content was much richer than what is captured by the three business  and  financial  knowledge  questions  included  in  our  follow‐up  survey.  In  particular,  the  course  20    discussed business practices, such as account keeping and use of bank accounts, the impact on  which we examine in the following section.     Finally, we test whether the training had a differential effect on individuals who had a business  at  baseline  and  who  did  not. 11   The  results  in  Column  3  of  Table  7  indicate  that  the  effect  is  slightly larger for individuals who owned a business at baseline than for individuals who did not,  but this difference is not statistically significant.     V.V. Evaluation Results – Effects on Business Outcomes  This  section  examines  the  effects  of  the  training  on  business  outcomes,  including  survival,  practices, and performance.    Business Creation and Survival  First, we study whether the training had an effect on business survival and business creation.  Table 8 includes our complete sample, i.e. all entrepreneurs who responded to the follow‐up  survey,  independent  of  whether  they  had  a  business  at  baseline  or  not.  We  find  that  the  training had no significant effect on whether our study participants had a business at follow‐up  (Column 1). This is true for individuals with below and above median financial literacy levels at                                                               11  As mentioned above, we did not stratify by this variable in the randomization, but the variable is balanced across  treatment and control groups.  21    baseline  (Column  2).  It  is  also  true  for  individuals  who  had  a  business  at  baseline  and  for  individuals who did not have a business at baseline (Column 3), implying that the training did  not  increase  the  likelihood  of  starting  a  business  among  potential  entrepreneurs.  In  fact,  our  data shows that only one new business started up in our sample during the study period. The  last two columns of Table 8 include only individuals who had a business at baseline in order to  examine whether the training promoted business survival. We do not find this to be the case.  Overall, we find no evidence that the training had an effect on business entry and survival.    Business Performance  The  remainder  of  this  section  analyzes  the  effect  of  the  training  on  business  outcomes  for  individuals who had a business at baseline and at follow‐up. We start by examining the impact  on business performance, as measured by one‐month profits. On average, the training did not  increase business profits (Column 1 of Table 9). However, the heterogeneous treatment effects  analysis  in  Column  2  suggests  that  the  training  increased  profits  by  KM  1,190  for  individuals  with  above  median  financial  literacy  at  baseline,  compared  to  an  average  of  KM  2,218  in  the  control group. 12  This effect corresponds to a 54 percent increase in profits. However, the effect  is only statistically significant at the 15 percent level, possibly because the profit data are noisy                                                               12  As a robustness check, Columns 3 and 4 of Table 9 display profits regressions with data winsorized at the 1%  level.  The  results  are  essentially  the  same  as  in  Columns  1  and  2,  implying  that  the  results  are  not  driven  by  outliers.  22    and because about one‐third of the clients did not provide profit data, reducing the sample size  to 108.    To supplement the profit data, we also asked business owners whether they had maintained,  increased, or decreased monthly profits compared to one year earlier. All entrepreneurs who  had a business at baseline and at follow‐up answered this question. Consistent with the results  in  Columns  1  through  4  of  Table  9,  the  last  two  columns  of  Table  9  show  that,  on  average,  entrepreneurs in the treatment group were not significantly more likely to have said that their  profits increased over the past year, compared to the control group. However, entrepreneurs  with above median financial literacy at baseline were 14.3 percent more likely than their peers  in the control group to have stated that their profits increased over the past year (compared to  a  base  of  18.9  percent  in  the  control  group).  However,  this  effect  is  also  only  statistically  significant  at  the  15  percent  level.  Overall,  the  evidence  in  Table  9  suggests  that  the  training  increased business profits for entrepreneurs with above median financial literacy at baseline by  (by 54 percent), although the results are not statistically significant at conventional levels.    Business Growth  Next,  we  examine  whether  the  training  promoted  business  growth  among  existing  firms.  We  consider  different  measures  of  business  size,  as  reported  in  Table  10.  First,  similar  to  our  question regarding profits, we asked entrepreneurs whether they had maintained, increased or  23    decreased  sales,  compared  to  one  year  ago.  As  with  profits,  we  do  not  find  a  statistically  significant  effect  of  the  training  on  whether  sales  increased  over  the  past  year,  on  average  (Column 1). However, entrepreneurs with above median financial literacy at baseline were 16.7  percent more likely to say that their sales increased over the past year than their peers in the  control group. This increase in equivalent to a doubling in the percentage of entrepreneurs who  said that their sales increased compared to one year ago, going from about 16 to 33 percent.     Our second measure of business size is number of employees, but we do not find a statistically  significant  effect  of  the  training  on  this  variable  (Columns  3  and  4).  Finally,  we  asked  respondents  whether  their  firm  expanded  its  installations  during  the  past  year.  As  shown  in  Columns 5 and 6, the training did not cause firms to expand their installations.     To  summarize,  the  results  in  Table  10  indicate  that  the  training  increased  sales  for  entrepreneurs with above median financial literacy at baseline, but we do not find an effect on  the more slow‐moving measures of firm growth, such as number of employees or expansion of  business installations. Such variables tend to be sticky and it is possible that changes would be  observed in the long‐run.     Business Practices and Investments  24    In  order  to  gain  a  better  understanding  of  the  channels  through  which  the  training  affected  business  decisions,  we  examine  the  impact  on  a  number  of  self‐reported  business  practices  (Table 11) and investments (Table 12). First, we find that entrepreneurs in the treatment group  are 22 percent less likely than entrepreneurs in the control group to use personal accounts for  their  business  (Column  1  of  Table  11).  This  effect  appears  to  be  equally  strong  for  entrepreneurs with below and above median financial literacy at baseline (Column 2). Second,  we  test  whether  the  training  had an  effect  on  using  credit  cards  for  the  business,  but  do  not  find this to be the case (Columns 3 and 4) 13 .     Next, Table 12 displays the effects of the training on a number of investments or changes that  the  entrepreneurs  report  to  have  made  in  their  businesses  during  the  past  year.  The  results  show that the training caused treatment group entrepreneurs to be 10.6 percent more likely to  invest their savings in the business than their peers in the control group (Columns 1 and 2). We  also  find  that  treatment  group  entrepreneurs  were  16.5  percent  more  likely  to  have  implemented  new  production  processes  than  control  group  entrepreneurs  (compared  to  a  mean  of  12  percent).  On  the  other  hand,  Table  12  does  not  show  a  significant  effect  of  the  training  on  developing  new  products  and  on  starting  new  marketing  campaigns.  As  a  final  measure, we compute RHS aggregated z‐scores for all outcome measures reported in this table,                                                               13  Note that we do not find any effects on keeping business accounts. However, our follow‐up data shows that the  proportion  of  businesses  in  both  treatment  and  control  groups  that  keep  accounts  is  very  high,  more  than  95  percent in each group.  25    following  the  methodology  in  Kling,  Leibman,  and  Katz  (2007).  These  results  are  reported  in  Columns  9  and  10,  and  show  that  the  aggregate  impact  on  business  investments  is  large,  positive, and statistically significant.      A notable finding in this analysis of business outcomes and practices is the difference in effects  of  the  training  on  individuals  with  below  and  above  median  financial  literacy  at  baseline.  We  find that both entrepreneurs with below and above median financial literacy changed some of  their  business  practices,  such  as  separating  personal  accounts  from  business,  and  making  investments  in  their  business;  however,  only  entrepreneurs  with  above  median  financial  literacy  at  baseline  reported  increases  in  sales  and  profits  as  a  result  of  the  training.  These  findings  suggest  that  baseline  knowledge  and  information  conveyed  in  the  training  act  as  complements in increasing the productivity and sales of a business.      V.VI. Evaluation Results – Treatment Effects on Loan Behavior  Adding  to  our  analysis  on  business  outcomes,  this  section  investigates  whether  the  business  training  program  changed  loan  behavior.  In  order  to  do  so,  we  analyze  very  detailed,  high  frequency administrative data from Partner. Since our sample may borrow from other sources  than Partner, we supplement the administrative data with a question on the firms’ overall loan  portfolio from the follow‐up survey.       26    We  start  by  examining  whether  the  training  had  an  effect  on  the  number  of  loans  taken  out  from Partner. As reported in Table 13, there is no statistically significant effect of the training  on the probability of taking out a loan from Partner in the post‐training period (Columns 1 and  2). Similarly, the training did not have an effect on the number of loans taken out (Columns 3  and 4). Finally, the treatment effect on the overall loan portfolio, that is having a business loan  from any source, is also negligible (Columns 5 and 6).     Next,  we  examine  the  treatment  effects  on  the  characteristics  of  new  loans  taken  out  from  Partner,  using  the  sample  of  loans  that  our  study  participants  took  out  after  the  training  (80  loans). The training did not significantly change the average loan amount (Columns 1 and 2 of  Table  14).  However,  we  detect  a  significant  treatment  effect  on  the  number  of  installments.  Specifically, the results in Columns 3 and 4 show that treatment entrepreneurs are more likely  to negotiate a larger number of installments than control group entrepreneurs. On average, the  training  increased  the  number  of  installments  from  22.7  to  27.6  (a  difference  of  about  5  months).  The  fact  that  treatment  group  entrepreneurs  tend  to  obtain  longer‐term loans  than  control group entrepreneurs is consistent with our finding from the previous section that they  tend  to  make  new  investments  in  their  businesses  (since  investment  loans  often  have  longer  terms than working capital loans). Finally, the treatment effect on the interest rate is negative,  but it is small and not statistically significant.    27    In Table 15, we examine loan default and restructuring. We find that the treatment effect on  loan  payments  being  past  due  and  loan  write‐off  is  negative,  but  not  statistically  significant  (Columns 1‐4). Note, however, that the average values of these variables in the control group  are very low, ranging from less than 1 to 6 percent, depending on how default is defined.     The significant finding from Table 15 is on loan restructuring. We find that the treatment group  is 3.4 percent more likely than the control group to refinance its loans with Partner (Column 4).  This is a large effect considering that only 4 percent of the control group refinanced its loans  with  Partner  during  this  period.  Hence,  the  treatment  almost  doubles  the  likelihood  of  refinancing loans. This refinancing typically takes the form of a lower interest rate or a longer  loan term.     Table  16  repeats  the  default  and  restructuring  analysis  with  heterogeneous  effects.  Here,  we  find  negative  treatment  effects  on  loan  default,  though  these  are  only  significant  at  the  15  percent level.     Overall,  the  loan  analysis  shows  no  impact  on  loan  amounts,  but  significant  impacts  on  loan  restructuring for existing loans and longer terms for new loans. The results on loan default are  weak, and show some negative impact for firms with low ex‐ante financial literacy.      28    VI. Conclusion  In  this  paper,  we  rigorously  test  the  impact  of  business  and  financial  training  for  young  entrepreneurs in Bosnia. We find that while the training program does not influence business  survival,  it  does  significantly  improve  business  practices  and  investments  among  surviving  businesses.  Specifically,  treatment  businesses  are  significantly  more  likely  to  implement  new  production  processes  and  to  inject  new  investment  into  the  business,  consistent  with  the  central  theme  of  the  training  which  was  to  encourage  more  capital  growth.  Further,  we  find  treatment  businesses  are  more  likely  to  separate  personal  and  business  accounts,  refinance  their loans for more favorable terms, and obtain new loans with lower repayment installments.     We  do  not  find  significant  average  treatment  effects  of  our  training  program  on  business  performance.  However,  we  identify  significant  heterogeneous  effects.  Specifically,  entrepreneurs with relatively high ex‐ante financial literacy exhibit improvements in sales due  to the training program. The effects on profits are also positive for this sub‐group, showing an  increase in profits due to the training by 54 percent, though only statistically significant at the  15 percent level.     Our  results  have  important  policy  implications  for  business  promotion  and  growth.  One  clear  message from our analysis is that lack of business knowledge is not the primary constraint to  new  entrepreneurship;  we  do  not  find  any  significant  impact  of  our  treatment  on  business  29    entry or exit. Hence, while programs aimed to promote new business start‐up should certainly  consider business training as part of their promotion package, this training should not be the  sole intervention. Related research has identified other much stronger constraints to business  development and growth, such as lack of capital (Bianchi and Bobba, 2010; De Mel, McKenzie  and Woodruff, 2008; and Gertler, Martinez, and Rubio‐Codina, 2006).     While business training does not impact the extensive margin, we show significant effects on  existing  entrepreneurs,  and  on  specific  aspects  of  their  businesses.  We  find  that  teaching  entrepreneurs  the  value  of  capital  investment  indeed  encourages  them  to  change  business  practices  that  allow  for  greater  innovation,  for  instance  by  implementing  new  production  processes and making personal investments in the business. These are encouraging results and  identify business training as an important policy tool to help improve outcomes for youth‐led  businesses.              30    References  Banerjee, Abhijit, Esther Duflo, Rachel Glennerster, and Cynthia Kinnan. 2010. “The Miracle of  Microfinance? Evidence from a Randomized Evaluation,” Working Paper.    Bianchi,  Milo  and  Matteo  Bobba.  2010.  “Liquidity,  Risk,  and  Occupational  Choices,”  Working  Paper.    Bloom, Nicholas, Aprajit Mahajan, David McKenzie, and John Roberts. 2010. “Why Do Firms in  Developing  Countries  Have  Low  Productivity?”  American  Economic  Review  Papers  &  Proceedings, 100(2): 619‐23.    Bruhn,  Miriam  and  Inessa  Love.  2009.  “The  Economic  Impact  of  Banking  the  Unbanked:  Evidence from Mexico.” World Bank Policy Research Paper No. 4981.    Bruhn, Miriam, and David McKenzie. 2009. "In Pursuit of Balance: Randomization in Practice in  Development  Field  Experiments."  American  Economic  Journal:  Applied  Economics,  1(4):  200– 232.    Bruhn,  Miriam,  Dean  Karlan,  and  Antoinette  Schoar.  2010.  “What  Capital  is  Missing  in  Developing Countries?”American Economic Review: Papers & Proceedings, 100 (2): 629‐33.  31      Cole,  S.,  and  N.  Fernando.  2008.  “Assessing  the  Importance  of  Financial  Literacy,”  Asian  Development Bank: Finance for the Poor.    Cole, Shawn, Thomas Sampson, and Bilal Zia. 2010. ”Prices or Knowledge? What Drives Demand  for Financial Services in Emerging Markets?” Forthcoming, Journal of Finance.     De Mel, Suresh, David McKenzie, and Christopher Woodruff. 2008. “Returns to Capital: Results  from a Randomized Experiment.” Quarterly Journal of Economics, 123(4): 1329‐72.    Demirgüç‐Kunt, Asli, Leora Klapper, and Georgios Panos. 2007. “The Origins of Self‐ Employment.” Washington, DC: Development Research Group, World Bank, February.    Gertler, Paul, Sebastian Martinez, and Marta Rubio‐Codina. 2006. “Investing Cash Transfers to  Raise Long Term Living Standards.” World Bank Policy Research Paper No. 3994.    Karlan,  Dean  and  Martin  Valdivia.  2010.  “Teaching  Entrepreneurship:  Impact  of  Business  Training  on  Microfinance  Clients  and  Institutions,”  Forthcoming,  Review  of  Economics  and  Statistics.    32    Klapper, Leora, Anamaria Lusardi, and Georgios Panos.  2010. “Financial Literacy and Financial  Crisis: Evidence from Russia,” Working Paper, The World Bank.      Kling,  Jeffrey,  Jeffrey  Liebman,  and  Lawrence  Katz.  2007.  "Experimental  Analysis  of  Neighborhood Effects," Econometrica, Vol. 75, Issue 1, 83‐119.     Levine, Ross, 2005. “Finance and Growth: Theory and Evidence,” in Philippe Aghion and  Steven Durlauf, eds: Handbook of Economic Growth (Elsevier Science).    Lusardi,  Annamaria,  and  Peter  Tufano.  2008.  “Debt  Literacy,  Financial  Experience  and  Overindebtedness,”  Working  Paper,  Dartmouth  College  and  Harvard  Business  School. 33    Appendix 1: Content of Business and Financial Education Training Courses    Day 1  Module 1: General Concepts (1 hour)  • What is entrepreneurship? – General knowledge, facts and ideas.  • Who is an entrepreneur? – General info about who is an entrepreneur  • Advantages and disadvantages of being an entrepreneur  • What are micro, small and medium enterprises?  • How to recognize a business opportunity?  • Types of business activities:   o Main business activity/source of income  o Secondary business activity/source of income  • Legal business types  o Independent businesses/sole proprietors (crafts, sales, services, etc...)  o Limited liability companies (LLC)  o Advantages of independent business and LLC  • To register or not? Steps for registering a business.  • Making investments in the business for it to grow  34    • Tax  system  in  Bosnia  and  Herzegovina.  What  is  VAT?  Difference  between  independent  businesses  and  mandatory  VAT  payers.  Examples  to  illustrate  how  VAT  works.  VAT  was  introduced in Bosnia in 2008, so many are still unfamiliar with how it works.     Module 2: Business Plan (2 hours)  • What is a business plan?  • Importance of business planning and a business plan  • Steps in developing a business plan  o  Analysis of current situation:  Internal organization   SWOT analysis   Team exercise to practice SWOT analysis for business type of their choice  o Defining business goals  Importance of business investment  Basics of marketing and market research  Basics of financial planning, projecting financial performance/income    Day 2    Module 3: Marketing (1 hour)  35    • What is marketing and why it is important for business?  • What is market? Supply and demand. Market research (size, potential, segmentation, etc. –  all in the context of the business plan)    • Customer behavior?  What is important to know about our buyers? How to communicate  through marketing?   • Marketing mix 4P   1. Product. Brand. Packaging.  2. Price. Sales strategies. Discounts, etc.  3. Promotion  4. Place (distribution)    Module 4: Understanding and Managing the Firm’s Finances (1 hr 30 min)  • What are finances? Basics of financial analysis as related to a business plan.   • Costs. What are costs? Types of costs. Managing and cost planning.  • Income and expenses and related planning.   • Keeping household finances separate from business income and expenses  • Basic financial reports. Balance sheet and income sheet.  • What is a cash flow? How to analyze cash flow for the needs of a business plan?    36    Module 5: Business Growth (30 min)  • What are investments and why are they important?   • Growth planning.  What is growth and what is business development? Internal and external  growth.  • How to grow healthy?  • Financing  a  growing  venture.    Internal  and  external  sources.    Personal  investments  and  Partnerships    • Final thoughts (for those doing only 5 modules)    Day 3     Module 6: The Importance of Financial Literacy in times of Financial Crisis (3 hours)  • Financing sources (pro’s and cons)  o Internal financing  o Loans & how to get them  Purchase of an investment or appreciable asset via debt as leverage  Upsides – greater returns, availability of funds, etc…  Downsides – risk, loss of investment, loan balance payment, etc…  Banks   Microcredit organizations  37    Family/Friends  o Government sources  Funds available at Municipality, Canton and Entity level  o Non‐governmental sources   o International & EU acceptance funds  • Importance of financial responsibility  o CRK (Central Credit Registry)  What it is and how it works  Credit consequences for failure to pay on time  Managing your credits & loans  • Interest rates  o Description of simple and compound interest  Compare bank interest rates and show matrix of potential returns  Basic formula to calculate simple & compound interest  o Rule of 70 or 72 (doubling shortcut)  o Common types of interest charged   Annual vs. effective interest rate  o Credit cards and interest  Interest on credit  Interest on cash  38    o Financial help resources  • Diversification  o Why diversity  Real life example i.e. selling umbrellas and sunscreen  o Diversification effects & return expectations  Smaller returns but smaller losses  Reduction in fluctuation of income  o Concept of correlation  Income from correlated vs. uncorrelated assets  Example  –  i.e.  investing  in  crops  whose  yield  depends  on  different  set  of  preconditions  Investing money in stock market vs. savings account deposits  o Diversification strategies  Spread  the  investment  portfolio  through  different  vehicles  –  in  this  case  different sources of income  By risk  By industry or geography  • Short & Long term  o Definition of short and long term investing in real assets & ventures  o Importance of seeing the entire picture  39    Do you have necessary information to make sound decisions  Compare your options on their true merits  o Why long term is more predictive of future performance    o Understanding periodic fluctuations in performance  o Defining investment goals  • The Devil’s in the Details  o Legal language  o Penalty clauses with loans  o Hidden fees  o Marketing traps  • Final thoughts  o Managing yours and expectations of others  o What can you fall back on?    40    Appendix 2: Survey Questions Measuring Financial Literacy and Business Knowledge    1. If  you  have  a  choice  to  invest  1,000  KM  with  one  of  three  friends  with  whom  would  you  invest?    Note,  there  is  a  possibility  your  investment  will  fail  and  you  would  lose  your  invested money.    1 Friend with an investment with highest return in the past month   2  Friend with an investment  with the highest return in the previous year   3  Friend with investment with low return and low risk   4 Invest a portion with all of them    997  Don’t know    2. Suppose you owe 1,000 KM on a loan from Partner and the interest rate you are charged is  20%  per  year  compounded  annually.  Compounding  means  that  interest  for  the  year  is  calculated  at  the  end  of  each  year  based  on  the  total  outstanding  amount,  inclusive  of  principal and interest.  If you didn’t pay anything off, at this interest rate, how many years  would  it  take  for  the  amount  you  owe  to  double?  Read  the  options  and  mark  the  box  in  front of the indicated answer.  1    2 years;  2    less than 5 years;  41    3    5 to 10 years;  4    more than 10 years;  5    Do not know.  6    Refuse to answer.    3. Suppose you owe 3,000 KM on a loan from Partner. You  pay a minimum payment of $30  each  month.  At  an  Annual  Percentage  Rate  of  12%  (or  1%  per  month),  how  many  years  would  it  take  to  eliminate  your  debt  if  you  made  no  additional  new  charges?  Read  the  options and mark the box in front of the indicated answer.  1    Less than 5 year;  2    Between 5 and 10 years;  3    Between 10 and 15 years;  4    Never, you will continue to be in debt;  5    Do not know.  6    Refuse to answer.    4. All individuals & legal subjects making less than 50,000 in taxable income are obligated to  pay VAT? Listen and mark the indicated response.          1  Yes         2  No            997  Do not know  42      5. Do you know what the Central Credit Registry is? Listen and mark the indicated response.          1  Yes         2  No    6. In  difficult  times  businesses  sometimes  seek  to  temporarily  lower  prices  in  hope  of  attracting  new  customers.    They  plan  to  increase  prices  at  a  later  day  when  market  conditions  improve.    If  price  of  a  product  is  100  KM  and  is  lowered  by  30%  how  many  percent  does  the  product  price  have  to  be  increased  by  to  return  to  the  original  100  KM  price. Read the options and mark the indicated response.  1    By 30%  2    Less than 30%   3    More than 30%  4    Do not know     7. Suppose  you  are  a  farmer  facing  unpredictable  market  conditions  where  prices  are  fluctuating. In order to best protect your income stream, you should: Read the options and  mark the indicated response.   1  Specialize in one crop                     2  Grow multiple crops for which historically prices have moved in the same direction  43     3  Grow multiple crops for which historically prices have moved in different directions    997 Do not know    8. Suppose  you  operate  a  farm  and  are  interested  in  purchasing  a  crop  processing  machine.  The machine costs 1,000KM. You do not have the resources to pay for the machine in cash  so  the  seller  offers  you  two  financing  options:  a)  Pay  12  fixed  monthly  installments  of  100KM  each;  b)  Borrow  $1,000KM  from  the  seller  for  a  12  month  loan  at  a  15%  annual  interest  rate.  Which  is  the  more  advantageous  offer?  Read  the  options  and  mark  the  indicated response.   1 Option (a)         3 They are the same   2 Option (b)         997  Do not know    44    Table 1: Dependent Variable: Interested in Participating in Training Program? This table reports the results from OLS regressions estimating which borrower and loan characteristics predict whether the client was interested in the business training course. These characteristics all come from Partner’s client database. Robust standard errors. *** indicates statistical significance at the 1% level, ** at the 5% level, * at the 10% level, and + at the 15% level. (1) (2) (3) Female -0.127∗∗∗ -0.125∗∗∗ -0.131∗∗∗ (0.025) (0.025) (0.025) Residence Status == Domiciled 0.012 0.013 -0.025 (0.037) (0.037) (0.041) Rural -0.027 -0.021 -0.013 (0.027) (0.027) (0.030) Ethnicity == Bosniak 0.052 0.049 0.061 (0.063) (0.063) (0.064) Age -0.047 -0.046 -0.053 (0.041) (0.041) (0.041) Age Squared 0.001 0.001 0.001 (0.001) (0.001) (0.001) Sector == Farming & Livestock -0.010 -0.005 0.005 (0.034) (0.035) (0.035) Sector == Services 0.011 0.013 0.021 (0.029) (0.029) (0.029) Loan Amount Outstanding -0.000 -0.000 (0.000) (0.000) Dummy Late Days of Payment in Current Loan ≥ 1 0.053∗∗ 0.057∗∗ (0.024) (0.024) Constant 1.067∗ 1.011∗ 1.646∗∗∗ (0.553) (0.554) (0.560) Branch FEs No No Yes R-squared 0.015 0.016 0.019 N 1783 1783 1783 Table 2: Baseline Characteristics This table reports summary statistics for the business loan clients included in an experiment on the impact of a comprehensive business and financial literacy program. Panel A describes the full sample consisting of 445 clients. Panel B describes the subsample of clients who had a business at baseline, consisting of 267 clients. The last column provides p-values for a difference-in-means test between the treatment and control groups. *** indicates statistical significance at the 1% level, ** at the 5% level, and * at the 10% level. Total Treatment N Control N p-value Panel A. Full Sample Demographics Age 445 28.138 297 28.041 148 .802 Rural 445 .714 297 .696 148 .697 Ethnicity == Bosniak 445 .97 297 .959 148 .574 Residence Status == Domiciled 445 .889 297 .851 148 .258 Completed Secondary School 444 .852 297 .803 147 .19 Risk Averse 445 .68 297 .709 148 .53 Stratification Variables Female 445 .35 297 .351 148 .98 Baseline Fin Lit Score 445 2.673 297 2.608 148 .63 Missing Profit in March 2009 445 .202 297 .209 148 .855 Sector == Farming & Livestock 445 .266 297 .27 148 .924 Sector == Services 445 .461 297 .466 148 .922 Panel B. Had a Business at Baseline Business Characteristics No. of Employees (incl. owner) 260 2.28 168 2.054 92 .562 Net Profits in March 2009 (KM) 229 1365.238 147 905.122 82 .302 Net Profits in March 2009 (KM), Winsorized Top and Bottom 1% 229 841.429 147 728.293 82 .381 Business Age (Months) 253 58.267 165 59.739 88 .823 Has Any Business Assets 267 .936 172 .916 95 .539 Registered 267 .203 172 .295 95 .093* Has Checking/Savings Account For Business 267 .483 172 .463 95 .762 Has Credit Line 267 .907 172 .916 95 .81 Has Credit Card 267 .07 172 .063 95 .837 Extends Trade Credit 255 .79 167 .807 88 .758 Accepts Trade Credit 249 .627 161 .67 88 .499 Keeps Business Accounts 267 .494 172 .6 95 .098* Has Participated In Other Financial Literacy Course 267 .087 172 .053 95 .306 Stratification Variables Female 267 .314 172 .337 95 .703 Baseline Fin Lit Score 267 2.692 172 2.663 95 .871 Missing Profit in March 2009 267 .163 172 .158 95 .917 Sector == Farming & Livestock 267 .198 172 .253 95 .299 Sector == Services 267 .541 172 .463 95 .226 Table 3: Predictors of Use of Financial Services This table reports the results of OLS regressions of the determinants of formal financial services usage and business practices. The first column for each dependent variable includes the full sample and the second column focuses exclusively on individuals who had a business at baseline. Robust standard errors. *** indicates statistical significance at the 1% level, ** at the 5% level, * at the 10% level, and + at the 15% level. (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Has Bank Has Bank Has Has Has Has Accepts Accepts Extends Extends Keeps Keeps Account Account Credit Credit Credit Credit Trade Trade Trade Trade Business Business (Check- (Check- Line Line Card Card Credit Credit Credit Credit Accounts Accounts ing or ing or Savings) Savings) Baseline Financial 0.057∗∗∗ 0.042∗∗ 0.019∗ 0.019+ 0.000 0.008 0.034∗ 0.033+ 0.031∗∗ 0.037∗ 0.041∗∗ 0.032+ Literacy Score (0.017) (0.021) (0.010) (0.012) (0.009) (0.014) (0.018) (0.022) (0.016) (0.021) (0.018) (0.022) Female -0.111∗∗ -0.149∗∗ 0.054∗ 0.062∗ -0.045∗∗ -0.033 -0.107∗∗ -0.132∗ -0.078∗ -0.039 -0.037 -0.020 (0.049) (0.062) (0.029) (0.035) (0.019) (0.030) (0.054) (0.071) (0.047) (0.064) (0.052) (0.066) Residence Status == 0.001 0.021 0.165∗∗∗ 0.152∗∗ -0.042 -0.000 0.143∗ 0.086 0.174∗∗ 0.150+ 0.047 0.079 Domiciled (0.068) (0.075) (0.058) (0.074) (0.042) (0.051) (0.074) (0.098) (0.071) (0.091) (0.071) (0.094) Rural -0.159∗∗∗ -0.073 0.065∗ 0.010 -0.032 -0.041 -0.075 -0.052 0.027 0.002 -0.070 -0.006 (0.051) (0.062) (0.035) (0.045) (0.028) (0.039) (0.054) (0.067) (0.046) (0.059) (0.053) (0.062) Ethnicity == Bosniak -0.001 -0.194+ -0.051 -0.018 -0.140 -0.252+ 0.045 0.026 -0.093 -0.117 -0.047 -0.040 (0.133) (0.125) (0.074) (0.121) (0.106) (0.166) (0.146) (0.161) (0.100) (0.109) (0.133) (0.134) Age 0.103 0.072 0.033 0.047 0.045∗ 0.010 0.043 -0.131 0.037 -0.045 -0.036 -0.090 (0.086) (0.118) (0.053) (0.081) (0.025) (0.042) (0.089) (0.112) (0.076) (0.097) (0.085) (0.110) Age2 -0.002 -0.001 -0.001 -0.001 -0.001∗ -0.000 -0.001 0.003 -0.001 0.001 0.001 0.002 (0.002) (0.002) (0.001) (0.001) (0.000) (0.001) (0.002) (0.002) (0.001) (0.002) (0.002) (0.002) Completed Secondary 0.182∗∗∗ 0.136∗ -0.015 -0.053 0.033+ 0.008 -0.054 -0.103 -0.085+ -0.067 0.095 0.009 School (0.058) (0.070) (0.038) (0.047) (0.021) (0.034) (0.069) (0.088) (0.058) (0.082) (0.068) (0.090) Has Participated in Other 0.341∗∗∗ -0.045 0.115 0.144 -0.087 0.108 Business Training (0.084) (0.084) (0.093) (0.103) (0.105) (0.098) Risk Averse -0.001 -0.005 -0.045 -0.101∗ 0.057 -0.116∗∗ (0.062) (0.040) (0.036) (0.061) (0.058) (0.058) Registered 0.369∗∗∗ -0.033 0.021 0.223∗∗∗ 0.085 0.485∗∗∗ (0.063) (0.051) (0.046) (0.070) (0.061) (0.056) Sector == Farming & -0.077 -0.007 0.008 0.178∗ 0.120 -0.064 Livestock (0.088) (0.049) (0.044) (0.096) (0.091) (0.093) Sector == Services -0.083 -0.041 0.017 0.178∗∗ 0.143∗∗ -0.110+ (0.071) (0.041) (0.043) (0.077) (0.070) (0.068) Has Any Business Assets 0.157+ 0.151∗ 0.033 -0.137 -0.099 0.147 (0.096) (0.090) (0.025) (0.150) (0.137) (0.126) Constant -1.038 -0.565 0.215 0.016 -0.362 0.153 -0.282 2.282+ 0.145 1.307 0.677 1.533 (1.168) (1.614) (0.727) (1.132) (0.331) (0.572) (1.216) (1.572) (1.051) (1.356) (1.155) (1.513) R-squared 0.090 0.253 0.047 0.024 0.022 0.038 0.022 0.068 0.023 0.008 0.017 0.201 N 444 267 444 267 444 267 407 249 414 255 444 267 Baseline Mean of Dep Var 0.446 0.476 0.905 0.910 0.059 0.067 0.582 0.643 0.778 0.796 0.457 0.532 Table 4: Predictors of Take Up This table reports the results of OLS regressions of the determinants of take up of the business training program. The first column for each dependent variable includes the full sample of treated individuals and the second column focuses on the subsample of treated individuals who had a business at baseline. Robust standard errors. *** indicates statistical significance at the 1% level, ** at the 5% level, * at the 10% level, and + at the 15% level. (1) (2) Attended Attended training training Baseline Financial -0.004 -0.011 Literacy Score (0.017) (0.024) Female 0.044 0.096 (0.054) (0.080) Residence Status == -0.007 0.065 Domiciled (0.078) (0.106) Rural -0.144∗∗ -0.115+ (0.059) (0.077) Ethnicity == Bosniak -0.284∗ -0.464∗∗∗ (0.168) (0.164) Age -0.096 -0.100 (0.093) (0.159) Age2 0.002 0.002 (0.002) (0.003) Completed Secondary -0.004 -0.061 School (0.071) (0.105) Has Participated in Other 0.045 Business Training (0.120) Risk Averse -0.036 (0.067) Registered 0.114 (0.092) Sector == Farming & -0.064 Livestock (0.110) Sector == Services -0.019 (0.092) Has Any Business Assets 0.137 (0.136) Constant 1.938+ 2.247 (1.276) (2.182) R-squared 0.025 0.046 N 297 172 Baseline Mean of Dep Var 0.219 0.227 Table 5: Business and Financial Knowledge This table reports business and financial knowledge at baseline, exit test, and follow up among the sample of follow up respondents. The exit test was administered after the training, and is thus available only for respondents who attended the training. The p-values reported in column 6 report the statistical significance of a paired mean-comparison test between the exit test and the baseline. The p-values reported in column 8 report the statistical significance of a paired mean-comparison test between the follow up and the baseline. *** indicates statistical significance at the 1% level, ** at the 5% level, * at the 10% level. N Baseline Exit Test Follow Up Exit Test − Baseline p-value Follow Up − Baseline p-value Panel A. Treatment Invited to Training Q1. Knows past returns doesn’t predict future returns 264 .375 Q2. Knows compound interest 264 .538 Q3. Knows making min. payments doesn’t eliminate debt 264 .057 Q4. Knows VAT law 264 .36 .402 .042 .25 Q5. Knows what the credit registry is 264 .193 .356 .163 0** Q6. Understands perecentage calculations 264 .519 Q7. Understands diversification 264 .17 .458 .288 0** Q8. Can compare financing options 264 .455 Total Score (all 8 questions) 264 2.667 Total Score (Q4 and Q5) 264 .553 .758 .205 0** Attended Training Q1. Knows past returns doesn’t predict future returns 112 .393 .33 -.062 .252 Q2. Knows compound interest 112 .527 .33 -.196 .002** Q3. Knows making min. payments doesn’t eliminate debt 112 .045 .063 .018 .482 Q4. Knows VAT law 112 .339 .625 .464 .286 0** .125 .022** Q5. Knows what the credit registry is 112 .241 .571 .429 .33 0** .188 0** Q6. Understands perecentage calculations 112 .527 .411 -.116 .085* Q7. Understands diversification 112 .17 .304 .446 .134 .008** .277 0** Q8. Can compare financing options 112 .366 .277 -.089 .15 Total Score (all 8 questions) 112 2.607 2.911 .304 .083* Total Score (Q4 and Q5) 112 .58 1.196 .893 .616 0** .313 0** Panel B. Control Q1. Knows past returns doesn’t predict future returns 132 .348 Q2. Knows compound interest 132 .53 Q3. Knows making min. payments doesn’t eliminate debt 132 .045 Q4. Knows VAT law 132 .371 .348 -.023 .614 Q5. Knows what the credit registry is 132 .212 .326 .114 .005** Q6. Understands perecentage calculations 132 .538 Q7. Understands diversification 132 .182 .455 .273 0** Q8. Can compare financing options 132 .439 Total Score (all 8 questions) 132 2.667 Total Score (Q4 and Q5) 132 .583 .674 .091 .146 Table 6: Financial Perception This table reports financial perception at baseline and exit test among the sample of follow up respondents. The exit test was administered after the training, and is thus available only for respondents who attended the training. The p-values reported in the last column report the statistical significance of a paired mean-comparison test between the exit test and the baseline. *** indicates statistical significance at the 1% level, ** at the 5% level, * at the 10% level. N Baseline Exit Test Exit Test − Baseline p-value Panel A. Treatment Invited to Training Thinks Financial are Skills Important in Business 264 .216 Risk Averse (Coin Toss) 264 .686 Strongly Agree/Agree w/: Not sure risky investment even if big possible profit 264 .451 Prefers to finance vehicle via credit 264 .542 Thinks good credit history can help obtain larger/better loans 264 .174 Attended Training Thinks Financial are Skills Important in Business 112 .205 .563 .357 0** Risk Averse (Coin Toss) 112 .679 .804 .125 .01** Strongly Agree/Agree w/: Not sure risky investment even if big possible profit 112 .384 .42 .036 .558 Prefers to finance vehicle via credit 112 .607 .464 -.143 .009** Thinks good credit history can help obtain larger/better loans 112 .223 .75 .527 0** Panel B. Control Thinks Financial are Skills Important in Business 132 .235 Risk Averse (Coin Toss) 132 .735 Strongly Agree/Agree w/: Not sure risky investment even if big possible profit 132 .417 Prefers to finance vehicle via credit 132 .538 Thinks good credit history can help obtain larger/better loans 132 .212 Table 7: Impact on Business and Financial Knowledge This table reports results from a randomized experiment measuring the impact of a comprehensive business and financial literacy program. The questions included in the dependent variable are the following: knows VAT law (Q53), knows what the credit registry is (Q55), diversification (Q59). The specification in column 1 is given by Y1i = T reatmenti + Stratai + Y0i + W ave2i + εi where i indexes individuals, Y1i refers to values at follow up, and Y0i refers to values at baseline. W ave2 is a dummy for the second wave of the follow up survey. Strata are defined by gender, financial literacy score at baseline ≥ 3 sector, and missing profits. Robust standard errors. *** indicates statistical significance at the 1% level, ** at the 5% level, * at the 10% level, and + at the 15% level. (1) (2) (3) Q53, Q55, Q53, Q55, Q53, Q55, Q59 Q59 Q59 Treatment -0.002 0.239∗ 0.045 (0.104) (0.131) (0.165) Treatment ∗ Above Median Baseline Financial Literacy -0.248 (0.181) Has Business at Baseline 0.105 (0.168) Treatment ∗ Has Business at Baseline 0.092 (0.205) Strata Dummies Yes Yes Yes Control for Baseline Outcome Yes Yes Yes Wave 2 Dummy Yes Yes Yes R-squared 0.188 0.183 0.186 N 396 396 396 Mean of Dep Var in Control Group 1.129 1.129 1.129 Mean of Dep Var for [Above Median Baseline Fin Lit/Has Business at Baseline] in Control Group 1.311 1.189 Mean of Dep Var for [Below Median Baseline Fin Lit/Did Not Have Business at Baseline] in Control Group 0.897 0.973 Table 8: Impact on Business Survival and Business Entry This table reports results from a randomized experiment measuring the impact of a comprehensive business and financial literacy program. The specification in columns 1 and 4 is given by Yi = T reatmenti + Stratai + W ave2i + εi where i indexes individuals. Columns 2, 3, and 5 add interaction terms. W ave2 is a dummy for the second wave of the follow up survey. Strata are defined by gender, financial literacy score at baseline ≥ 3, sector, and missing profits. Robust standard errors. *** indicates statistical significance at the 1% level, ** at the 5% level, * at the 10% level, and + at the 15% level. (1) (2) (3) (4) (5) Has Business Has Business Has Business Business Sur- Business Sur- at Follow up at Follow up at Follow up vived vived Treatment -0.015 0.021 0.030 0.021 0.063 (0.053) (0.079) (0.027) (0.062) (0.099) Treatment ∗ Above Median Baseline Financial Literacy -0.063 -0.074 (0.107) (0.128) Has Business at Baseline 0.645∗∗∗ (0.055) Treatment ∗ Has Business at Baseline -0.004 (0.068) Constant -0.001 -0.019 -0.270 0.489 0.506 (0.045) (0.057) (0.279) (0.364) (0.378) Strata Dummies Yes Yes Yes Yes Yes Wave 2 Dummy Yes Yes Yes Yes Yes R-squared 0.089 0.089 0.409 0.115 0.117 N 396 396 396 267 267 Mean of Dep Var in Control Group 0.439 0.439 0.439 0.611 0.611 Mean of Dep Var for [Above Median BL Fin Lit/Has Business at BL] in Control Group 0.500 0.611 0.500 Mean of Dep Var for [Below Median BL Fin Lit/Did Not Have Business at BL] in Control Group 0.362 0.000 0.362 Table 9: Impact on Business Performance This table reports results from a randomized experiment measuring the impact of a comprehensive business and financial literacy program. The specification in columns 1, 3, and 5 is given by Y 1i = T reatment i + Strata i + W ave2 i + Y 0i + M issingY 0i + ε i, and columns 2, 4, and 6 add interaction terms. i indexes individuals, Y 1i refers to values at follow up. Y 0i refers to values at baseline, and missing values are replaced with zero. M issingY 0i is a dummy equal to 1 if Y 0i is missing and was replaced with zero. W ave2 is a dummy for the second wave of the follow up survey. Strata are defined by gender, financial literacy score at baseline ≥ 3, sector, and missing profits. Robust standard errors. *** indicates statistical significance at the 1% level, ** at the 5% level, * at the 10% level, and + at the 15% level. (1) (2) (3) (4) (5) (6) Net Profits Net Profits Net Profits Net Profits Increased Increased May 2010 May 2010 May 2010 May 2010 profits profits winsorized winsorized top and top and bottom 1% bottom 1% Treatment -65.828 -1485.965 -163.190 -1500.938 0.051 -0.102 (837.310) (1149.150) (780.597) (1149.533) (0.080) (0.129) Treatment ∗ Above Median Baseline Financial Literacy 2675.040+ 2519.111+ 0.245+ (1635.066) (1556.320) (0.162) Constant 2377.752∗∗ 2677.872∗∗ 2331.718∗∗ 2622.510∗∗ 0.103 0.168 (1155.323) (1156.483) (1094.439) (1080.078) (0.193) (0.192) Strata Dummies Yes Yes Yes Yes Yes Yes Wave 2 Dummy Yes Yes Yes Yes Yes Yes Control for BL Outcome and Dummy for Missing BL Outcome Yes Yes Yes Yes No No R-squared 0.179 0.202 0.189 0.212 0.084 0.099 N 108 108 108 108 170 170 Mean of Dep Var in Control Group 2642.162 2642.162 2642.162 2642.162 0.224 0.224 Mean of Dep Var for Above Median Baseline Financial Literacy in Control Group 2217.500 2217.500 0.189 Mean of Dep Var for Below Median Baseline Financial Literacy in Control Group 3141.765 3141.765 0.286 Table 10: Impact on Business Growth This table reports results from a randomized experiment measuring the impact of a comprehensive business and financial literacy program. The specification in columns 1, 3, and 5 is given by Y 1i = T reatment i + Strata i + W ave2 i + Y 0i + M issingY 0i + ε i, and columns 2, 4, and 6 add interaction terms. i indexes individuals, Y 1i refers to values at follow up. Y 0i refers to values at baseline, and missing values are replaced with zero. M issingY 0i is a dummy equal to 1 if Y 0i is missing and was replaced with zero. W ave2 is a dummy for the second wave of the follow up survey. Strata are defined by gender, financial literacy score at baseline ≥ 3, sector, and missing profits. Missing baseline outcome values have been replaced with zero, and regressions that control for the baseline outcome include a dummy for missing baseline outcome. Robust standard errors. *** indicates statistical significance at the 1% level, ** at the 5% level, * at the 10% level, and + at the 15% level. (1) (2) (3) (4) (5) (6) Increased Increased Log num of Log num of Expanded Expanded sales sales employees employees installations installations in past year in past year Treatment 0.062 -0.115 -0.011 -0.116 -0.024 -0.002 (0.079) (0.127) (0.120) (0.239) (0.074) (0.113) Treatment ∗ Above Median Baseline Financial Literacy 0.282∗ 0.169 -0.035 (0.158) (0.274) (0.148) Constant 0.095 0.170 0.027 0.075 -0.018 -0.027 (0.194) (0.191) (0.301) (0.327) (0.213) (0.221) Strata Dummies Yes Yes Yes Yes Yes Yes Wave 2 Dummy Yes Yes Yes Yes Yes Yes Control for Baseline Outcome and Dummy for Missing Baseline Outcome No No Yes Yes Yes Yes R-squared 0.067 0.087 0.306 0.308 0.233 0.234 N 169 169 170 170 170 170 Mean of Dep Var in Control Group 0.207 0.207 0.681 0.681 0.276 0.276 Mean of Dep Var for Above Median Baseline Financial Literacy in Control Group 0.162 0.657 0.324 Mean of Dep Var for Below Median Baseline Financial Literacy in Control Group 0.286 0.724 0.190 Table 11: Impact on Business Practices This table reports results from a randomized experiment measuring the impact of a comprehensive business and financial literacy program. The specification in columns 1 and 3 is given by Y 1i = T reatment i + Strata i + W ave2 i + Y 0i + M issingY 0i + ε i, and columns 2 and 4 add interaction terms. i indexes individuals, Y 1i refers to values at follow up. Y 0i refers to values at baseline, and missing values are replaced with zero. M issingY 0i is a dummy equal to 1 if Y 0i is missing and was replaced with zero. W ave2 is a dummy for the second wave of the follow up survey. Strata are defined by gender, financial literacy score at baseline ≥ 3, sector, and missing profits. Missing baseline outcome values have been replaced with zero, and regressions that control for the baseline outcome include a dummy for missing baseline outcome. Robust standard errors. *** indicates statistical significance at the 1% level, ** at the 5% level, * at the 10% level, and + at the 15% level. (1) (2) (3) (4) Uses personal Uses personal Has credit card Has credit card account for account for for business for business business business Treatment -0.218∗∗∗ -0.278∗∗ -0.000 -0.022 (0.079) (0.137) (0.063) (0.095) Treatment ∗ Above Median Baseline Financial Literacy 0.095 0.034 (0.166) (0.129) Constant 0.517∗∗ 0.542∗∗ 0.056 0.065 (0.217) (0.219) (0.059) (0.067) Strata Dummies Yes Yes Yes Yes Wave 2 Dummy Yes Yes Yes Yes Control for Baseline Outcome and Dummy for Missing Baseline Outcome Yes Yes Yes Yes R-squared 0.287 0.289 0.146 0.147 N 169 169 170 170 Mean of Dep Var in Control Group 0.655 0.655 0.172 0.172 Mean of Dep Var for Above Median Baseline Financial Literacy in Control Group 0.676 0.189 Mean of Dep Var for Below Median Baseline Financial Literacy in Control Group 0.619 0.143 Table 12: Impact on Business Investements This table reports results from a randomized experiment measuring the impact of a comprehensive business and financial literacy program. The specification in columns 1, 3, 5, and 7 is given by Y 1i = T reatment i + Strata i + W ave2 i + Y 0i + M issingY 0i + ε i, and columns 2, 4, 6, and 8 add interaction terms. i indexes individuals, Y 1i refers to values at follow up. Y 0i refers to values at baseline, and missing values are replaced with zero. M issingY 0i is a dummy equal to 1 if Y 0i is missing and was replaced with zero. W ave2 is a dummy for the second wave of the follow up survey. Strata are defined by gender, financial literacy score at baseline ≥ 3, sector, and missing profits. Missing baseline outcome values have been replaced with zero, and regressions that control for the baseline outcome include a dummy for missing baseline outcome. Robust standard errors. *** indicates statistical significance at the 1% level, ** at the 5% level, * at the 10% level, and + at the 15% level. (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Invests sav- Invests sav- Developed Developed Implemented Implemented Started new Started new Average z Average z ings in busi- ings in busi- new prod- new prod- new pro- new pro- marketing marketing score score ness ness ucts in past ucts in past duction duction campaign in campaign in year year processes in processes in past year past year past year past year Treatment 0.106∗∗ 0.087∗ 0.067 0.091 0.165∗∗∗ 0.156∗∗ 0.002 -0.068 0.398∗∗∗ 0.321∗ (0.044) (0.047) (0.064) (0.093) (0.061) (0.077) (0.059) (0.093) (0.119) (0.176) Treatment ∗ Above Median Baseline Fin Lit 0.029 -0.039 0.014 0.110 0.122 (0.079) (0.127) (0.115) (0.118) (0.235) Constant -0.053 -0.046 0.210 0.199 -0.202∗∗ -0.198∗∗ 0.230 0.260 -0.144 -0.112 (0.048) (0.046) (0.220) (0.232) (0.095) (0.100) (0.209) (0.209) (0.359) (0.358) Strata Dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Wave 2 Dummy Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Control for Baseline Outcome and Dummy Yes Yes Yes Yes Yes Yes Yes Yes No No for Missing Baseline Outcome R-squared 0.190 0.190 0.243 0.244 0.227 0.227 0.131 0.137 0.242 0.243 N 169 169 170 170 170 170 170 170 170 170 Mean of Dep Var in Control Group 0.017 0.017 0.155 0.155 0.121 0.121 0.121 0.121 0.000 0.000 Mean of Dep Var for Above Median Baseline 0.027 0.189 0.162 0.108 0.064 Financial Literacy in Control Group Mean of Dep Var for Below Median Baseline 0.000 0.095 0.048 0.143 -0.113 Financial Literacy in Control Group Table 13: Loan Outcomes This table reports results from a randomized experiment measuring the impact of a comprehensive business and financial literacy program. Data in columns 1 to 4 come from Partner’s administrative loan data, data in columns 5 to 6 come from the follow up survey. W ave2 is a dummy for the second wave of the follow up survey. Strata are defined by gender, financial literacy score at baseline ≥ 3, sector, and missing profits. OLS regressions at the client level, with robust standard errors. Took out loan ex post is a dummy equal to one if a client ever took out a loan after December 2009. Number of loans taken out ex post is equal to zero if the client never took out a loan ex post. The sample in the last two columns consists of respondents who had a business at both baseline and follow up. *** indicates statistical significance at the 1% level, ** at the 5% level, * at the 10% level, and + at the 15% level. (1) (2) (3) (4) (5) (6) Took out Took out Num of Num of Currently Currently loan ex-post loan ex-post loans taken loans taken has loan for has loan for out ex post out ex post business business Treatment 0.002 -0.006 0.019 0.016 -0.045 -0.118 (0.037) (0.055) (0.040) (0.059) (0.074) (0.120) Treatment ∗ Above Median Baseline Financial Literacy 0.014 0.004 0.118 (0.075) (0.080) (0.152) Constant 0.159∗∗ 0.164∗∗ 0.187∗ 0.189∗ 0.565∗∗ 0.581∗∗ (0.076) (0.077) (0.099) (0.098) (0.261) (0.254) Strata Dummies Yes Yes Yes Yes Yes Yes Wave 2 Dummy No No No No Yes Yes Control for Baseline Outcome No No No No Yes Yes R-squared 0.053 0.054 0.051 0.051 0.188 0.191 N 445 445 445 445 170 170 Mean of Dep Var in Control Group 0.169 0.169 0.169 0.169 0.759 0.759 Mean of Dep Var for Above Median Baseline Financial Literacy in Control Group 0.171 0.171 0.757 Mean of Dep Var for Below Median Baseline Financial Literacy in Control Group 0.167 0.167 0.762 Table 14: Partner Loan Data This table reports results from a randomized experiment measuring the impact of a comprehensive business and financial literacy program. Data come from Partner’s administrative loan data. Strata are defined by gender, financial literacy score at baseline ≥ 3, sector, and missing profits. OLS regressions at the client-loan level. The sample in these regressions consists only of new loans that were disbursed starting January 2010. Robust standard errors clustered at the client level. *** indicates statistical significance at the 1% level, ** at the 5% level, * at the 10% level, and + at the 15% level. (1) (2) (3) (4) (5) (6) Loan Loan Num of in- Num of in- Nominal int Nominal int amount amount stallments stallments rate rate Treatment 0.429 603.030 4.939∗ 7.866∗∗∗ -0.117 0.205 (849.055) (1421.171) (2.865) (2.918) (0.645) (0.581) Treatment ∗ Above Median Baseline Financial Literacy -1006.045 -4.886 -0.537 (1760.976) (5.268) (1.165) Loan amount 0.002∗∗∗ 0.002∗∗∗ -0.000∗∗∗ -0.000∗∗∗ (0.001) (0.001) (0.000) (0.000) Constant 5499.571∗∗∗ 4896.970∗∗ 7.394 4.627 22.705∗∗∗ 22.400∗∗∗ (1549.940) (1931.229) (6.142) (5.895) (1.370) (1.329) Strata Dummies Yes Yes Yes Yes Yes Yes R-squared 0.200 0.205 0.537 0.544 0.452 0.453 N 80 80 80 80 80 80 Mean of Dep Var in Control Group 4392.000 4392.000 22.680 22.680 20.461 20.461 Mean of Dep Var for Above Median Baseline Financial Literacy in Control Group 4500.000 23.429 21.383 Mean of Dep Var for Below Median Baseline Financial Literacy in Control Group 4254.545 21.727 19.288 Table 15: Partner Loan Data This table reports results from a randomized experiment measuring the impact of a comprehensive business and financial literacy program. Data come from Partner’s administrative loan data. Strata are defined by gender, financial literacy score at baseline ≥ 3, sector, and missing profits. OLS regressions at the client-loan-month level. The sample in these regressions consists only of loans that are active starting January 2010. Robust standard errors clustered at the individual level. *** indicates statistical significance at the 1% level, ** at the 5% level, * at the 10% level, and + at the 15% level. (1) (2) (3) (4) (5) More than More than Num of days Given up or Refinanced 30 days past 60 days past past due written off or restruc- due due tured Treatment -0.019 -0.008 -1.993 -0.002 0.034∗ (0.018) (0.013) (1.838) (0.003) (0.020) Loan amount -0.000∗∗ -0.000∗∗ -0.000∗∗∗ -0.000∗∗∗ 0.000 (0.000) (0.000) (0.000) (0.000) (0.000) Constant 0.064∗∗ 0.050∗∗ 7.819∗∗ 0.005 0.034 (0.030) (0.022) (3.274) (0.005) (0.055) Strata Dummies Yes Yes Yes Yes Yes Month Dummies Yes Yes Yes Yes Yes R-squared 0.041 0.039 0.050 0.015 0.090 N 3901 3901 3901 3901 3901 Mean of Dep Var in Control Group 0.060 0.035 6.285 0.006 0.039 Table 16: Partner Loan Data This table reports results from a randomized experiment measuring the impact of a comprehensive business and financial literacy program. Data come from Partner’s administrative loan data. Strata are defined by gender, financial literacy score at baseline ≥ 3, sector, and missing profits. OLS regressions at the client-loan-month level. The sample in these regressions consists only of loans that are active starting January 2010. Robust standard errors clustered at the individual level. *** indicates statistical significance at the 1% level, ** at the 5% level, * at the 10% level, and + at the 15% level. (1) (2) (3) (4) (5) More than More than Num of days Given up or Refinanced 30 days past 60 days past past due written off or restruc- due due tured Treatment -0.048+ -0.028 -5.095+ -0.007+ 0.015 (0.032) (0.022) (3.293) (0.004) (0.032) Treatment ∗ Above Median Baseline Financial Literacy 0.052 0.036 5.589+ 0.009∗ 0.034 (0.037) (0.026) (3.837) (0.005) (0.040) Loan amount -0.000∗∗ -0.000∗∗ -0.000∗∗ -0.000∗∗∗ 0.000 (0.000) (0.000) (0.000) (0.000) (0.000) Constant 0.082∗∗ 0.062∗∗ 9.757∗∗ 0.009+ 0.046 (0.035) (0.026) (3.885) (0.006) (0.056) Strata Dummies Yes Yes Yes Yes Yes Month Dummies Yes Yes Yes Yes Yes R-squared 0.044 0.042 0.054 0.015 0.091 N 3901 3901 3901 3901 3901 Mean of Dep Var in Control Group 0.060 0.035 6.285 0.006 0.039 Mean of Dep Var for Above Median Baseline Financial Literacy in Control Group 0.038 0.021 3.989 0.003 0.028 Mean of Dep Var for Below Median Baseline Financial Literacy in Control Group 0.086 0.052 9.035 0.010 0.052